dcegm

Submodules

• dcegm.asset_correction
• dcegm.backward_induction
• dcegm.check_func_outputs
• dcegm.cli
• dcegm.egm
• dcegm.final_periods
• dcegm.interfaces
• dcegm.interpolation
• dcegm.law_of_motion
• dcegm.likelihood
• dcegm.numerical_integration
• dcegm.pre_processing
• dcegm.simulation
• dcegm.solve_single_period
• dcegm.toy_models

Classes

setup_model

Package Contents

class dcegm.setup_model(model_config: Dict, model_specs: Dict, utility_functions: Dict[str, Callable], utility_functions_final_period: Dict[str, Callable], budget_constraint: Callable, state_space_functions: Dict[str, Callable] = None, stochastic_states_transitions: Dict[str, Callable] = None, shock_functions: Dict[str, Callable] = None, alternative_sim_specifications: Dict[str, Callable] = None, debug_info: str = None, model_save_path: str = None, model_load_path: str = None, use_stochastic_sparsity: bool = False)

model_specs

specs_without_jax

model_config

model_funcs

model_structure

batch_info

params_check_info

income_shock_draws_unscaled

income_shock_weights

solve(params, load_sol_path=None, save_sol_path=None)

Solve a discrete-continuous life-cycle model using the DC-EGM algorithm.

Parameters:

• params (pd.DataFrame) – Params DataFrame.

• options (dict) – Options dictionary.

• utility_functions (Dict[str, callable]) – Dictionary of three user-supplied functions for computation of: (i) utility (ii) inverse marginal utility (iii) next period marginal utility

• budget_constraint (callable) – Callable budget constraint.

• state_space_functions (Dict[str, callable]) – Dictionary of two user-supplied functions to: (i) get the state specific feasible choice set (ii) update the endogenous part of the state by the choice

• final_period_solution (callable) – User-supplied function for solving the agent’s last period.

• transition_function (callable) – User-supplied function returning for each state a transition matrix vector.

solve_and_simulate(params, states_initial, seed, load_sol_path=None, save_sol_path=None)

Solve the model and simulate it.

Parameters:

• params – The parameters for the model.

• states_initial – The initial states for the simulation.

• wealth_initial – The initial wealth for the simulation.

• n_periods – The number of periods to simulate.

• seed – The random seed for the simulation.

• alt_model_funcs_sim – Alternative model functions for simulation.

Returns:

A dictionary containing the solution and simulation results.

get_solve_func()

Create a fast function for solving that is jit compiled in the first call.

get_solve_and_simulate_func(states_initial, seed)

Create a fast function for solving and simulation that is jit compiled in the first call.

create_experimental_ll_func(params_all, observed_states, observed_choices, unobserved_state_specs=None, return_model_solution=False, use_probability_of_observed_states=True, slow_version=False)

validate_exogenous(params)

get_state_choices_idx(states)

Get the indices of the state choices for given states.

get_child_states(state, choice)

get_child_states_and_calc_trans_probs(state, choice, params)

Get the child states for a given state and choice and calculate the transition probabilities.

get_full_child_states_by_asset_id_and_probs(state, choice, params, asset_id, second_continuous_id=None)

Get the child states for a given state and choice and calculate the transition probabilities.

compute_law_of_motions(params)

get_n_state_choices_per_period()

solve_partially(params, n_periods, return_candidates=False)

set_alternative_sim_funcs(alternative_sim_specifications, alternative_specs=None)